typesense

typesense: Build Fast, Typo-Tolerant Search Experiences

Typesense is a C++ search engine for adding low-latency, typo-tolerant search to applications. Use it when you want a straightforward, self-hostable alternative to hosted search services, with options for filtering, faceting, vector search, and clustering.

View on GitHub Compare
Stars
26.6k
Forks
987
Stars, last 30 days
+208

Our take

Health score

55/100

How it is scored
Activity
55
Community
66
Issues
57
Pull requests
42

+208 stars in the last 30 days

Project history

Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.

Selected snapshot

Stars

27,132

Forks

1,031

Change since Nov 14, 2026: +1 stars·+1 forks

Overview

Typesense is a search engine that indexes application data and exposes search through an HTTP API. It aims to make fast, typo-tolerant search easier to build and operate, with sensible defaults and a self-contained server rather than a complex search platform to tune.

It suits teams building product search, site search, or discovery features that need filtering, ranking, and modern vector or semantic search capabilities. You can run it on your own infrastructure or use Typesense Cloud. It is not a general-purpose analytics or log-storage platform.

Key Features

  • Handles typographical errors in search queries.
  • Supports filtering, faceting, sorting, grouping, and distinct results.
  • Searches across multiple collections in a single request and supports query-time joins.
  • Provides geospatial search and sorting.
  • Supports vector, semantic, and hybrid search, including embeddings generated with supported models.
  • Offers conversational search and natural-language query understanding.
  • Includes scoped API keys, synonyms, and result curation controls.
  • Supports Raft-based clustering for high availability.

Use Cases

  • E-commerce teams can build product search with facets, filters, price sorting, and curated results.
  • Developers can add typo-tolerant search to a website, documentation portal, or application without relying on a hosted search service.
  • Teams with embeddings can build semantic, similarity, or hybrid keyword-and-vector search over their own data.
  • Applications with location-aware discovery can search and sort records by distance or geographic bounds.
  • Operators who need availability across multiple nodes can use Typesense's clustering support.

Getting Started

Run the server with Docker:

docker run -p 8108:8108 -v/tmp/data:/data typesense/typesense:29.0 --data-dir /data --api-key=YourApiKey

Then create a collection, index documents, and query them through the API or an official client. See the README and documentation for setup and API details.

Alternatives

  • paradedb: ParadeDB embeds full-text and vector search in PostgreSQL, while Typesense runs as a separate search engine with its own API.
ProjectLanguageLicenseStarsStatus
typesenseC++GPL-3.026.6kNot checked yet
paradedbRustAGPL-3.09.3kActive

Considerations

Typesense is an in-memory search engine, so memory use grows with indexed data and depends on the fields being indexed. Plan capacity around your dataset and query workload. Self-hosting means operating the server yourself; managed hosting is available separately through Typesense Cloud. The server uses GPL-3.0, so review the license before distributing modified versions. The repository reports 910 open issues, which may be useful context when assessing project support and outstanding work.

Found this useful?

Share it with someone who would like typesense.

Comparisons